Diameter Variation in Cellulose Fibrils A. J. BAILEY1 AND R. M. BROWN University of Minnesota, St. Paul, Minn.
Various plant fibers were disintegrated by mechanical means alone and by beating after different chemical treatments. Measurements of fibril diameters were made and analyzed statistically. The analysis showed that cellulose fibrils exist as uniform and true morphological units. Diameters of these fibrils are in the range of 0.928 to 0.956 p and are apparently independent of botanical origin, chemical treatment, or physical processing. The fact that cellulose fibrils fluctuate around a common average diameter, the amount of fluctuation being essentially the same for various species and treatments, has farHE degree of subdivision of a solid is of extreme importance in relation to chemical and colloidal changes. Kot infrequently physical appearance and properties are directly dependent upon particle size. Cellulose is markedly of this type. Changes in cellulose particle size by commercial beating processes are so complex that no means exists for accurately defining the change analytically or for exactly duplicating a given commercial beating operation. A tendency of cellulose to form particles of a definite size and resist further subdivision would be important in both physical and chemical processing. Cellulose esterification, for example, is often considered to be a micellar reaction in which some zones in a single micelle may be fully esterified or completely unchanged (18). Similarly, higher nitrates are known to exist on the fiber exterior, but lower nitrates persist in the interior (8). The present study had for its purpose the determination of the diameters of cellulose fibrils from different sources, of different chemical histories, and subjected to different mechanical processing, to learn the following: whether fibril diameters tend to be of an average size; how these sizes varied from the mean; how fibril diameters compared with diameters observed by other investigators; and what industrial importance might attach to fibrils which resisted further subdivision.
T
Experimental History Different fibril diameters have been reported by independent investigators. Balls and Hancock (6) concluded that cotton fibrils were 0.4 p in diameter, Jancke (14) in Herzog’s laboratory observed a width of 0.3 to 0.5 p , and Frey-Wyssling recorded 0.4 p (13). I n contrast, Freudenberg ( l a ) , Bailey and Kerr (4), Anderson and Moore ( d ) , and Anderson and Kerr (1) expressed the opinion that fibrils have no con1
Present address, University of Washington, Seattle, Wash.
reaching industrial significance. Two physical materials were found in all fibersa compact, dense, unit fibril and a substance forming a hydrogel upon comminution in water. The unit fibril was incapable of gelatinization in water by ordinary methods. The disintegration of fibers into a gel and an inert fraction indicates that freeness and similar tests have only limited and empirical significance. The observed course of fiber disintegration explains many of the complex physical and chemical changes accompanying commercial beating. Unit fibrils from various sources are illustrated by photomicrographs.
sistent size and grade down to the limits of microscopic resolution. Seifriz and Hock (16) recognized “primary fibrils” of 1.4 p diameter and “secondary fibrils” of 0.1-0.3 p diameter. Farr and Eckerson (10) reported a diameter of approximately 1.1 p for ellipsoidal particles and believed that they were a more or less constant building unit. Another investigator (S), who recorded a somewhat larger fibril diameter, believed that fibrils had a more or less uniform minimum diameter, or multiple of this diameter if grouped. As a result he proposed the term “unit fibril” to designate a fibril having this minimum uniform diameter and visually estimated the range to be from 3 to 4 p. I n the present investigation this diameter range was determined accurately and found to be from 0.93 to 0.96 p. Because the cylindricoids (3)formed from fibrils are approximately equidiniensional, the height is also in the range of 1.0 to 1.5 p ; the exact dimension requires statistical study. I n the present investigation unit fibrils could not be differentiated from Farr’s fibrils composed of ellipsoids. Although most of these earlier studies included only cotton, a few dealt with wood fibers. I n spite of marked difference in the type and development of fibers, there is no apparent reason why cellulose deposition cannot be identical in seed hairs and wood tracheids. A comparison of methods used to determine fibril diameters indicates that the fibril diameters actually found are correlated with the methods used. Generally speaking, the diameters obtained indirectly by inference from observations of zones in swollen sections of the walls have been smaller than those obtained by other methods. The larger diameters were usually obtained by simpler treatments, such as comminution in water. Because of the interference of Liesegang effects and pressure artifacts, there might be some danger in broadly inferring that zones or markings in drastically swollen material indicate fibril size. Little is known of the specificity of the gel involved. Varying
58
IKDUSTRIAL AND ENGINEERING CHEMISTRY
concentrations of reacting ions might bring about progressive peptization and coagulation with attendant reticulation and articulation. This might easily account for zones of varying density observed by some investigators. Only recently Barr (9) showed that some zones on transverse sections, believed by some workers to be growth rings, are in reality artifacts caused by pressure on the cover glass above swollen material. It is perhaps significant that many of the workers who separated and measured diameters of individual fibrils, rather than inferring diameter from wall markings or zones, have reported the higher diameters. T o define material in terms of analytical procedures is not new or unusual. Cellulose, for example, is designated as alpha, beta, gamma, Cross and Bevan, and holocellulose, depending on the method of preparation; yet an absolutely pure cellulose, giving theoretical yields of glucose, has yet to be prepared. It is entirely possible (a) that mild treatment might yield fibrils of a uniform size and (b) that more drastic swelling treatments might reduce these fibrils t o fibrils of smaller diameter by overcoming intermicellar attraction, but would not decrease their length since this could be brought about only by hydrolysis (shortening of chains). It is therefore rational to define fibrils not only on the basis of botanical origin, but also by chemical and physical histories. It might easily be the case that all recorded diameters have some significance, that agreement merely necessitates exact analytical definitions. Obviously, any clarification of such an important but contradictory and confusing concept would be of value. I n the interests of clarity, the term “fibril” is used here, as it appears in the literature, to mean a threadlike filament of varying length and diameter. The term “unit fibril” is used to indicate fibrils of definite diameter (about 1.0 p), whose length is still unknown. It is believed that fiber laminations are built u p of unit fibrils, that fibrils of greater diameter than 1.0 p are aggregates of unit fibrils, and that fibrils of diameters less than 1.0 p are strips torn longitudinally from unit fibrils.
Preparative and Observational Technique Some fibrous materials were obtained in various stages of industrial processing; others had no treatment. By processing these in the laboratory chemically or physically, or both, it was possible to eliminate the effects of a given beating apparatus or specific chemical treatment. A description and history of the various materials used follows: 1. Surgical cotton (Gossypium sp.), fiber len th reduced with scissors t o permit beating without roping and fumping, pebblemilled 2 hours (all beating was in distilled water only). 2. Commercial unbleached aspen soda pulp (Populus tremuloides), pebble-milled 1.5 hours. 3. Commercial unbleached aspen soda pulp, bleached overnight in a large excess of 5 per cent sodium hypochlorite solution, .. pebble-milled- 1 hour. 4. Commercial unbleached aspen soda pulp, pebble-milled 2 hours, hydrolyzed 2 hours with boiling, . . - 2 -per cent sodium hydroxide. 5. Commercial strawboard, pebble-milled 2 hours. 6. Southern pine (Pinus sp.) commercial kraft . paper, pebble. . milled 2 hours. 7. Southern pine kraft paper, bleached (similarly to sample 3) in 5 per cent sodium hypochlorite, pebble-milled 2 hours. (Both samples 6 and 7 were unbleached kraft paper which had had the usual beating and sizing. The paper was disintegrated in water before beating in this study.) 8. Thin western white pine ( Pinus monticola) shavings, pebblemilled 2 hours. 9. Western white pine shavings pulped with nitric acid and potassium chlorate, pebble-milled 2 hours. 10. Western white pine shavings pulped with nitric acid and potassium chlorate, bleached in 5 per cent sodium hypochlorite, (similarly to sample 3) pebble-milled 1 hour. ~
~
VOL. 32. NO. 1
11. Commercial Fir-Tex (Douglas fir, Pseudotsuga tarifolia) pulp, pebble-milled 2 hours. 12. Commercial Fir-Tex pulp, bleached in 5 per cent sodium
hypochlorite (similarly to sample 3), pebble-milled 2 hours. 13. Commercial Masonite (southern pine) pulp, pebblemilled 3 hours. 14. Commercial Masonite pulp, bleached in 5 per cent sodium hypochlorite (similarly to sample 3), pebble-milled 3 hours. 15. Western hemlock (Tsuga heterophylla) shavings, pulped with nitric acid and potassium chlorate, pebble-milled 2 hours. 16. Sitka spruce (Picea sitchensis) shavings, pulped with nitric acid and potassium chlorate, pebble-milled 2 hours.
Samples 2, 3, and 4 were from the same commercial pulp, and samples 6 and 7 were from the same paper. Lap pulp and paper were boiled a few minutes in dilute alkali, the water was squeezed out, and the mass was washed and kneaded in the hands t o loosen the fibers. The pulp was then disintegrated with a motor-driven stirrer and thickened by filtering. The pebble mill consisted of a n 800-ml. porcelain jar about half full of approximately half-inch pebbles and was driven by a synchronous motor a t 130 r. p. m. The mill was furnished by adding 100 ml. of distilled water and 2 to 3 grams (air dry) of fibrous material. All of the beating periods produced effects greatly in excess of commercial beating. All pulps were neutralized (if necessitated by prior treatment), or washed with sulfurous acid if oxidized, and then washed to neutrality before beating. Slides were prepared from diluted suspensions, mounted in a nonaqueous medium of high index of refraction, and observed with a n achromatic-aplanatic condenser of 1.40 N. A. (numerical aperture), an achromatic 1.9-mm. oil immersion objective of 1.30 N. A., and a filar micrometer with an achromatic eyepiece. The magnification was 930 diameters and the theoretical resolution slightly bett’er than 0.2 p. Measurements were made independently by an unprejudiced observer. Readings from the filar micrometer drum, calibrated by a stage micrometer, were rounded off t o the nearest 0.1 p, in spite of a resolution of 0.2 p. Readings were made in different parts of the slide where individual fibrils were separated and therefore could be measured accurately so that a representive sample was obtained and the effects of selective deposition on the slide were eliminated. The appearance of the beaten material on slides is shown in Figures 1 to 5. The resolution in all of these illustrations is less than half that obtained in making the measurements, owing to the necessity of obtaining optical contrast by diffraction. Variations in diameter of a single unit fibril are due to clots of gel, or to lying in a different focal plane; the result is the well known exaggeration of size of near objects by all short-focus lenses. These photomicrographs were prepared from airmounted fiber slides with no attempt t o place all fibrils in one plane. The distortion caused by a lens of less than 2-mm. focal length is increased 640 to 1080 times by the magnification. Bailey presented a complete description of the beaten material (3) ; he observed that the frayed fibers seemed to be composed of fibrils whose diameters were too uniform at various places in a single fibril and in different fibrils to be accidental. Figure 1 illustrates this condition. It was suspected that this uniformity indicated a morphological and fundamental unit of cellulose. These uniform fibrils are termed “unit fibrils” in the present discussion.
Statistical Analysis To determine the number of diameter measurements necessary t o obtain a reliable estimate of the average fibril diameter, a preliminary sample of sfty fibrils was obtained from each of the following sources-surgical cotton, unbleached aspen soda pulp, and hydrolyzed unbleached aspen soda pulp. The
JANUARY, 1940
FIQURE1 (above). SAMPLE 2, COMMERCIALUNBLEACHED ASPEN SODA PULP The fiber is just losing structural inte rity, releasin unit fi%& and gel. &ote the uniformity of fihrif diameters, the separated unit fibrils, and the cylindricoids ( x 640). FIQWRE 4 (below). Snnap~~ 16, SPRUCE PULPEDWITH NITUICACID AND POTABB I U M CBMRATLTI] ( x 1080)
INDUSTRIAL AND ENGINEERING CHEMISTRY
59
60
INDUSTRIAL AND ENGINEERING CHEMISTRY
averages, standard deviations, and standard errors of the averages are as follows:
-
Source of Fibril
Av. Diam.
Surgical ootton Pebble-milled unbleached aspen soda pulp Hydrolyzed unbleached aspen soda pulp
0.92
StandaTd Deviation Microns 0.059
Standard Error
TABLE11. ANALYSIS OF VARIANCE OF FIBRIL DIAMETER FROM CELLULOSES PULPED BY DIFFERENTPROCESSES
\
0.008
1.08
0.110
0.015
1.00
0.077
0.011
Source of Variation Between sources of ce11u 1ose Within sources (error) Total
These standard errors (0.008, 0.015, and 0.011) show that a sample of fifty fibrils gave an average with a sampling error much less than the power of resolution, 0.2 p, of the microscope. Even for the most variable of these sources-pebblemilled unbleached aspen-the sampling error is only one twentieth of this power of resolution. Because this high degree of precision is not necessary, the size of sample required to give a sampling error equal to one fifth of the power of resolution for the most variable source was computed by the formula for the standard error of the average as follows:
VOL. 32, NO. 1
Degrees of Sum of Freedom Squares 15 144
0.19 0.99
159
1.18
-
-
Mean Square 0.013 0.007
7-F TestActual F 5% 1% 1 . Q G 1.7
2.1
Just signidcant.
(Table 11) indicate that as a group the variation between the averages differs by more than could reasonably be expected if the sixteen samples had been drawn from the same fundamental population of unit fibrils. This general conclusion is indicated by the fact that the actual value of F, 1.9, is just a trifle larger than the value, 1.7, that occurs by chance five times out of one hundred trials (the level of significance arbiSD 2 0 1 1 2 N = trarily chosen by statisticians as the dividing line between (0.04) = (CfG) = sampling and true differences). F is the number of times one measure of variation is larger than the accidental or where N is the number of observations required to give an chance variation. average with a precision equal to one fifth the power of resoTo discover which of these averages might be grouped tolution, and SD is the standard deviation of pebble-milled gether and Considered typical of a fundamental population of aspen soda pulp. Although a sample of eight would satisfy cellulose unit fibrils, or conversely, which averages could not this arbitrary requirement, ten measurements were actually be considered members of such a population, all the possible used in order to simplify computations. differences between pairs of averages were subjected to the The average diameters of the fibrils from sixt’een samples t test (11, 19). The number of times an actual difference is drawn from nine different sources of cellulose treated, with greater than a sampling difference is designated as t . For one exception, by one or more of five different pulping procthe size of sample analyzed, the values of t that would occur ewes are given in Table I. five times in a hundred trials and only once in a hundred trials are 2.0 and 2.6. These figures multiplied by the standard error of a difference TABLE I. AVERAGEDIAMETERS OF FIBRILS FROM CELLULOSES give the significant and highly significant difTREATED BY DIFFERENT PROCESSES ferences with which the actual differences may Standard Diam. Av. Deviation Standrtrd Error Av. Of be compared. The standard error of a difference Source of Cellulose Fibrils between two means computed from samples of Micronsten is computed from the mean square for error Not Significantly Different from Fundamental Group in Table I1 by the following formula: Western hemlock shavings, pulped with HNOs and KClOa, 7 -
pebble-milled 2 hr. Commercial Fir-Tex pulp, bleached in 5 % NaOC1, pebblemilled 2 hr. Commercial strawboard, pebble-milled 2 hr. Surgical cotton Western white pine shavings pulped with HNOa and KClOa, Febble-milled 2 hr. Southern pine kraft paper, bleached in 5% NaOC1, pebblemilled 2 hr. Sitka spruce shavings, pulped with “Os and KClOs, I pebble-milled 2 hr. Thin western white pine shavings. pebble-milled 2 hr. Southern pine kraft paper, pebble-milled 2 hr. Commercial Masonite (southern pine) pulp, pebble-milled 3 hr Commkrcial unbleached as en soda pulp, bleached in 5 % NaOCl pebble-milled 1 gr. 0s and Western ’white pine shavings ulped with “ KClOa,,bleaphed in 5 % NaOCf pebble-milled 1 hr. Commercial Fir-Tex,pulp. pebble-milled 2 hr. Commercial Masonite pulp bleached In 5 % NaOCI, pebble-milled hr. Av. (based on 2samples of ten)
0.91
0.100
0,032
0.92 0.92 0.92“
0.082
0.026
0,094 0.093
0.030 0.011
0.93
0.066
0.021
0.94
0.047
0.015
0.94 0.95 0.95 0.96
0.047 0.055 0.100 0.047
0.015 0.017 0.032 0.015
0.96
0.082
0.97 0.96
00.066 .082
0.97 0,942
0.066 0.077
Very Highly Significantly Different from Fundamental Group Commercial unbleached aspen soda pulp hydrolyzed 2 hr. with NaOHunbleached a t looo C.. aspen pebble-milled 2 hr.pebble-milled 1.01b 0.080 Commercial soda pulp 1.5 hr. 1.07b 0.106 0 Based on 70 observations. b Based on 60 observations; all others based on 10 observations.
The primary question to be answered by these data is as follows: May these averages be considered estimates of the average fibril diameter of the same fundamental population of cellulose unit fibrils, or do they differ by more than could be expected if they had been drawn from such a population? To determine this, these data were analyzed by Fisher’s variance method (11) and Snedecor’s F test (19). The results
Standard error of a difference = 4 2 (mean square for error) =
d v
10
0.037
Any difference equal to a t least twice this standard error of a difference, or 0.074 p, would be considered a significant difference (i. e., more 0.026 than a sampling difference); and a difference 2.6 0.021 0.026 times this, or 0.096 p, a highly significant difference (i. e., one still less likely to occur as a 0.021 result of sampling). Therefore, all averages 0.007 between 0.91 and 0.07 or 0.98 p may be considered as belonging to the same population of 0.010 fibrils. As Table I shows, this group includes 0.013 all sources and treatments, except pebble-milled aspen soda pulp and hydrolyzed soda aspen pulp, with averages over 1 p , This group may therefore be considered the fundamental population of typical cellulose unit fibrils. If this group is assumed to represent the fundamental population of cellulose unit fibrils, the limits within which the trueaveragediameterof this population will fall may be computed from the average and standard error of this average. As Table I shows, the average diameter of the unit fibril is 0.942 1.1 * 0.007. Therefore the limits within which the true average falls with a probability of being correct ninety-five out of one hundred trials are this
JANUARY, 1940
INDUSTRIAL AND ENGINEERING CHEMISTRY
average plus and minus twice this standard error, or 0.928 to 0.956 p, Furthermore, not only because these treatments cover the range of severity of treatment of industrial conditions, but also because most of them are more severe, i t is felt that these limits do represent the probable range of the average diameters of the cellulose unit fibril from most sources. The variation of the individual fibrils from their averages is also shown in Table I by the standard deviations. For the fundamental group, the standard deviations range from 0.047 to 0.100 ,u. A Chi-square test (19) of this variation indicates, however, that these values are not significantly different. Therefore it appears that cellulose fibrils from most of these sources not only fluctuate around a common average diameter but also that the amount of this fluctuation is the same. The standard deviation of the fundamental group (Table I) is 0.077 p. Figure 6 gives the distribution of fibrils, based on the sixteen samples of ten each, from all the sources. If the two sets of significantly different aspen data are eliminated, the range will be reduced from 0.8-1.3 to 0.8-1.1 p. To determine whether or not the alkali-treated aspen soda pulp (KO. 4) and the unbleached aspen soda pulp could be considered as extreme members of this family of unit fibrils, a supplementary t test was used to compare the averages from the larger samples of data available for this material with the average, 0.942 p, of the group. For this test seventy measurements were available for unbleached pebble-milled aspen and sixty measurements for the alkali-treated aspen. The test was therefore more sensitive. The results of this analysis not only corroborated the original findings, but indicated that i t was highly improbable that these two sets of aspen fibril diameters were members of the fundamental group. A similar test was also made on the seventy measurements available for cotton. This more precise test indicated that the average diameter of the cotton fibrils was just significantly different from the average of the group. However, because a variation of this magnitude may occur five times in a hundred trials by chance alone, the cotton fibrils may be looked upon as smaller members of this fundamental group. Although it has been demonstrated by these statistical tests that some of the differences and measures of variation are statistically different, from a practical industrial standpoint these differences are without significance.
Optical, Chemical, Mechanical, and Botanical Limitations As noted above, optical resolution was 0.2 p. Average vision would require a magnification of 350 to 700 diameters to distinguish this separation; actually 930 diameters was used. An additional allowance was made for resolution less than this amount in calculating the number of observations necessary for a given accuracy. No claim is made that any individual measurement is more accurate than 0.2 p, although for mathematical reasons drum readings rounded off to the nearest 0.1 p were retained. Chemically, considerable diversity was obtained. Some fibers were wet only with water throughout the entire process from the block of wood to the slide. Others had alkaline or oxidative pulping and bleaching. Drastic nitric acid pulping and alkaline bleaching would presumably give nearly an ultimate oxycellulose product, visibly demonstrated by increased ease of disintegration in beating, shorter beating time, and friability of the pulp. The mildest treatment consisted only of agitation in water, whereas the most severe treatments were close to causing more or less complete degradation and disappearance of the cellulose as such; therefore we believe that not only was the range of severity of industrial conditions included, but that, for practical purposes,
61
the extreme limits of chemical conditions were approached. It was not intended to include all industrial materials but rather the range of severity of industrial conditions. Because of this great diversity there is little possibility that the formation of unit fibrils was a result of specific chemical action. The mechanical treatments used also varied greatly. From shavings to unit fibrils, some fibers were subjected bnly to pebble milling, of which Figure 2 is an example. The Masonite fibers had been defibered by steam explosion, the Fir-Tex had been hammer-milled, the straw had been rod-milled, the soda and kraft pulps had been through the macerating action of blowing the digester, and the kraft had been processed by standard beating tackle. The formation of unit fibrils by specific mechanical action is thus also remotely improbable. Botanically, cotton, straw, one hardwood, and four coniferous woods Tvere examined. RIorphologically these include tracheids, true fibers, seed hairs, and the heterogeneous cellular system of the grasses. Although no claims are made beyond the sources studied, it is believed that these samples embrace sufficient plant species to indicate thitt the occurrence of cellulose in unit fibrils is more or less general.
Fundamental Significance and Practical Applications The implications of uniform fibril diameter are extremely far-reaching. Probably the most important relate to industrial processing. It is obvious that a dense, resistant, morphological unit of cellulose (the unit fibril) is of immense importance in all physical processing of cellulose, such as beating. Resistant units (unit fibrils) are likewise important in the solution processes, such as cuprammonium and viscose, in that the unit fibril offers greater resistance t o solution than the interfibril material, or, as Farr showed recently (Q), in that fragments of unit fibrils resist cuprammonium solution entirely and form merely a microscopically visible part of a disperse system. The resistance of the dense unit fibrils is probably related to the relative amounts of the fibril fraction and gel fraction present, whereas the viscosity of the dispersion is also derived directly from these quantities and the specific viscosity of each. Similar considerations apply to the cellulose derivative processes such as nitrate and acetate. In addition, the unit fibril, which completely resists comminution in water and solution in cuprammonium alike, undoubtedly gives rise t o heterogeneous fibrillar reactions and thus adds to the difficulty of obtaining the fully nitrated or otherwise completely reacted cellulose derivative (probably measured directly in the longer time of reaction to secure contact and the higher concentration of reagents necessary t o accelerate diffusion into the unit fibril). Comminution in water demonstrated that unit fibrils retained structural integrity after all other fiber characteristics had been demolished. Undoubtedly this indicates greater cohesion due to secondary valence between adjacent micelles or chain molecules. That these forces are high is shown by the resistance of fibrils to disintegration. Apparently water alone did not satisfy intermicellar or intermolecular attraction sufficiently to lower it below the mechanical forces. Conceivably, but not probably, other bonds between micelles (chemical bonds from micelle to micelle) might account for this high resistance. Chemically resistant fibrils undoubtedly give rise to heterogeneous fibrillar reactions in a manner similar to that depicted by Sisson for micellar reactions (18),which consist of reacted outer layers and progress inward through lower derivatives to the unreacted core. It is possible that unit fibrils represent cellulose crystals as Sisson (17) suggested of Farr’s particles. Interior walls of cellulose fibers show spiral striations when swollen and bruised; that further mechanical demolition
62
INDUSTRIAL AND ENGINEERING CHEMISTRY
separates such a structure into closely packed unit fibrils lying side by side is visually demonstrated by Figure 1. The appearance of a clear gel, simultaneously with the separation of unit fibrils, points rather definitely to the existence of two different materials in the cell wall-one the physically resistant and chemically less active unit fibril, and the other the gel-fortning material apparently functioning as a cement between f i l d s and laminations ( 3 ) . This structure explains many of the changes occurring in beating, such as freeness, the development of strength, change in hydrolysis number, 70 7
2
60-
W 2
503
g
40-
LL
0
30-
V >
5
20-
0.8
0.9 FIBRIL
1.0
1.1
1.2
1.3
DIAMETER-MICRONS
FIGURE 6. DISTRIBUTION OF CELLULOSE FROM SIXFIBRILDIAMETER MEASUREMENTS TEEN SAMPLES OF TEN EACH,TAKENFROM KINE DIFFERENT SOURCES AND TREATED, WITH ONEEXCEPTION, BY ONE OR MOREOF FIVEDIFFERENT PULPING PROCESSES
the impossibility of completely gelatinizing cellulose by only mechaiiical means, and other associated phenomena. These are substantially the same conclusions reached by Brauns and Lewis (6’)on a highly purified pulp containing 99.1 per cent alpha-cellulose with practically no pentosan or uronic acid present. 13eating failed to hydrate this pulp, freeness did not change, and sheet-forming properties failed to appear, although the viscosity was fairly high. Also, fibers of this pulp treated with cuprammonium solution separated easily into individual lamellae, as though the original cementing material had been removed. In view of wall striations, the isolation of unit fibrils is direct evidence of strong and weak spots in the wall, in contrast to the homogeneous-lamella concept of some investigators. The presence of a gel-forming material and another substance, which does not gelatinize in beating, demonstrated by two independent investigations (8, 6), is highly important in the development and control of beating operations. Both of these materials would affect the viscosity of the material. Brauns and Lewis showed that the resistant cellulose may have high viscosity, yet no sheeting properties a t all. Further, it is apparent that viscosity may err widely if used to follow beating or predict strength proptrties. On slides prepared in this study, some fibers were covered by clots of a clear gel illustrated earlier (3). This was removed by immersing uncovered slides in sulfuric acid and hydrolyzing for several hours. The hydrolysis was so successful that the cement (gel) holding the fiber fragments to the slide was dissolved and the glass emerged clean. Partial hydrolysis was obtained by placing a few drops of the dilute acid on the slide and heating over a flame. Because the acid concentrated by evaporation, the threshold conditions to hydrolyze the gel without dissolving the unit fibrils were not apparent. After this hydrolysis many fibrils were still fastened to the slide in spots. It appeared that the gel had clotted the fibers and cemented them during drying, and that the
VOL. 32, NO. 1
gel was later hydrolyzed by the acid without apparent change to the unit fibrils except a sharpening and clarifying of the microscopic image. Similar results were obtained by alkaline hydrolysis with sodium hydroxide. Further work on the character, quantity, and location of this gel is contemplated. Farr’s observations of ellipsoidal particles forming a fibril are undoubtedly significant. Following the course of carbohydrate condensation on the cell wall, it would appear tenable that production of cellulose from the cytoplasm would soon cause saturation. According to Campbell’s theory of cellulose hydration (Y), longer chains would precipitate first and would be deposited preferentially on the growing particle due to their greater surface and difficulty of retaining a “hydrated atmosphere”. It is not difficult to understand how fibrils might be laid down with smaller particles or shorter chains filling the gaps and forming the matrix. It is possible that matrix material may be modified-for example, oxidation of an alcohol group to carboxyl forming a uronic acid. This explanation fits exactly the observations of Brauns and Lewis that their highly purified pulp had a uronide matrix. It is striking that two independent investigations have demonstrated particles, resolved from fibrils, of almost identical size. These have been termed “ellipsoid particles” (IO) and “cylindricoids” (S), but irrespective of name they appear to be derived from the same structure. Since the measurements recorded in this paper were made on dry fibrils, all that is necessary to concur with Farr’s diameter of 1.1 p is to add the amount of swelling caused by water (10 to 15 per cent) which then gives 1.1 p as the water-swollen diameter. The obvious explanation of the similarity in length of these particles is that sufficient ends of micelles occur a t one point to cause a weak spot in the fibril. That such a wide range of plants yields uniform fibrils is decidedly interesting. This would suggest that cellulose formation and condensation are conditioned only by chemical considerations and probably are merely expressions of identical equilibria in many, if not all, cellulose-producing plants; as such, they take precedence over the later function and location of the cell in the plant. Several investigators (1, 4, I S ) have reported what appears to be an anastomosing system of fibrils, the size of fibrils reportedly diminishing to invisibility. These observations have been made on more or less integral lamellae or fibers, usually after drastic swelling treatment. The “acid test” of anastomosing fibrils would appear to be actual isolation, which has never been accomplished. The complete absence, by any observer, of forked isolated fibrils, the uniformity of particle diameter observed ; the general uniformity of unit fibril diameter throughout the entire length, the general absence of tapering ends of fibrils (which would be present if torn from an anastomosing structure), and the 230 p length of some isolated fibrils (16) would appear to offer convincing evidence that fibrils are unbranched filaments. It seems probable that transverse sections of fiber walls containing extremely tenuous zones, obtained by drastic swelling tecliniques and susceptible to the complications of progressive peptization and coagulation, should be interpreted as effects of chemical action or combined chemical and mechanical treatment, rather than as structural features, if and until fibrils of this size are actually isolated, and by less objectionable technique The fact that two independent investigations (S,6) have demonstrated a pulp fraction which is completely incapable of being gelatinized by comminution in water, in contrast to a swelling treatment which may peptize cellulose completely, is reasonably conclusive evidence that the action of the two processes is different in kind. The concentric ring appearance of etched or swollen transverse sections and the apparent absence of radial markings on
JANUARY, 1940
INDUSTRIAL AND ENGINEERING CHEMISTRY
these rings have led some workers (I,,!+) to conclude that each lamella is continuous and integral. Therefore they formulated the hypothesis that fibrils formed by rupture of what seemed to be a homogeneous sleeve were of random size, in spite of the visibility of more or less uniform longitudinal markings in longitudinal sections. The error in concluding that each lamella is continuous would appear t o arise from two considerations: (a) that the importance of these longitudinal markings was overlooked and due explanation for the absence of the corresponding markings in transverse sections was not insisted upon; and ( b ) that the gaps between fibrils in a single lamella, when viewed in transverse section, slanted sufficiently from the axis of vision so that the optical contrast was below that required for threshold visibility. Undoubtr edly responsible for this continuous appearance of lamellar cross sections, then, are the rather large angle of fibril orientation from the vertical, the nearly perfect optical homogeneity of the specimen, the impossibility of obtaining any but a refraction and diffraction image, and the presence of peptizing agents, peptization or degradation products, and a partially peptized mother specimen; all contribute to fusion of refractive index and thus make optical differentiation and resolution impossible. Undoubtedly much remains to be learned about fiber laminations. In cotton the number of laminations appears to be related to diurnal light and temperature changes (2) yet wood fibers and tracheids, fully protected from light and partially protected from temperature changes by sereral or more inches of bark, show almost identical laminar and fibrillar structure. It would appear tenable that the laminations might be primarily due to diurnal changes in the rate of metabolism and thus cause a corresponding change in the rate of carbohydrate condensation on the cell wall. There is no evidence to indicate that a given lamination in a wood fiber is deposited on a certain day or time of day. Undoubtedly hysteresis exists between formation and deposition. How much is unknown, but it seems probable that thickness of laminations and unit fibril diameters are controlled by chemical equilibria. In other words, light and temperature effects in the leaves would have about as much influence on fibril diameter as the melting of snow fields a t the headwaters of a river would have on the change in water level or the size of waves a t the river mouth. I n conclusion, it should be repeated that different chemical treatments might explain small size differences, and that hydrolysis and oxidation are conducive to smaller diameters by reducing the amount of gel coated on the surfaces of the fibrils. This is undoubtedly the explanation of the high value for pebble-milled aspen soda pulp which has no other treatment. If we neglect the overwhelming statistical proof of unit fibrils for the purposes of discussion, it would still appear far from reasonable that, from a possible size range of about 0.1 to 10 p (i. e., 100 to 1) an unprejudiced observer should be able, merely by visual memory, t o select several hundred fibrils not varying from one another by more than 1 per cent of this range. It is by no means implied that smaller fibrils do not exist, but it is believed that the unit fibril is a natural morphological unit, probably associated with all plant cellulose. It is possible that chemical treatment, alone or prior t o physical treatment, might subdivide unit fibrils into the smaller particles reported by some investigators (1, 2, 4, 5, 13-16). Finally, the most significant finding of this study is not the exact dimension of the unit fibril, but the evidence that such a unit exists, with the manifold contingent implications, such as the presence of a gel, the nature of the physical disintegration of fibers, fibrillar reactions, viscosity effects, changes in freeness and hydrolysis number, development of strength, and other associated phenommt
63
Conclusions
1. Cellulose unit fibrils appear to have one more or less common and constant diameter, 0.9 to 1.0 p , independent of origin, chemical treatment, or mechanical processing. 2. At least two physical materials occur in the fiber wall, one a cementing gel, the other a compact fibril incapable of being gelatinized in water by ordinary methods. 3. Fiber laminations have alternately strong and weak spots; the distance from center to center of either is of the order of one micron. 4. Zone markings on drastically swollen material are a doubtful basis from which to infer fibril dimensions. 5. Anastomosing of fibrils appears to be extremely improbable. 6. Following the beating operation by viscosity, freeness, or any other single property is empirical and fundamentally inaccurate. 7 . A mechanism is provided for the explanation of fibrillar reactions, fiber disintegration, and the numerous physical and chemical changes which accompany this disintegration. 8. By inference, true wall laminations would appear to be a t least one micron thick, and thicker if heavily coated with gel, Acknowledgment The authors desire to express their appreciation to Phillip T. Schneider for making the micrometric observations.
Literature Cited (1) Anderson, D. B., and Kerr, T., IND. ENG.CHEM.,30, 48-54
(1938). (2) Anderson, D.G.,and Moore, J. H., Am. J . Botany, 24, 503-7 (1937). (3) Bailey, A. J.,Paper Ind., 17,735-9 (1936). (4) Bailey, I. W., and Kerr, T., J . ArnoZd Arboretum, 16, 273-300 (1935). (5) Balls, W. L.,and Hancock, H. A., Proc. Roy. SOC.(London), B93, 426-40 (1922). (6) Brauns, F. E.,and Lewis, H. F., Paper Trade J., 105, 35-7 (Sept. 2, 1937). (7) Campbell, W.B., Ibid., 100,35-8 (Feb. 14, 1935). (8) Desmaroux. J., and Mathieu, M., Compt. rend.. 192. 235-6 (1931). (9) Farr, W. K.,Contrib. Boyce Thompson Inat., 10, 71-112 (1938). (10) Farr, W.K., and Eckerson, S. H., Ibid., 6,189-203 (1934). (11) Fisher, R. A., “Statistical Methods for Research Workers”, 6th ed., Edinburgh, Oliver and Boyd, 1936. (12) Freudenberg, K., J. Chem. Education, 9,1171 (1932). (13) Frey-Wyssling, A., “Die Stoffausscheidung der hoheren PflanZen”, 1935. (14) Herzog, R.O.,Papier-Fabr., 23,121-2 (1925). (15) Ritter, G.J.. Paper Ind., 16,178-83 (1934). (16) Seifriz, W., and Hock, C. W., Paper Trade J., 102,36-8 (May 7, 1936). (17) Sisson, W. A., ContTib. Boyce Thompson Inst., 9,381-95 (1938). (18) Sisson, Tv. A.,IND.ENQ.CHEM.,30,530-7 (1938). (19) Snedecor, G.W.,“Statistical Methods”, Ames, Iowa, Collegiate Press, 1937.
Correction-Utilization
of Industrial Wastes
In this paper which appeared on pages 1323-30 in Kovember, 1939, we find an error on page 1324, second column. I t is stated that p-cymene is used in the manufacture of paint remover t o the extent of 750,000 gallons annually. We now learn through other sources that not over 20,000 gallons of p-cymene are recovered annually and that its chief outlet is in the manufacture of carvacrol, although at one time the proposal was made to use p-cymene for the purpose first stated. It was later found uneconomical to do so. H. E. HOWE F. J. VAX ANTWERPEN
